Convert List To Float List | Akatz

Convert List To Float List is a type-fixer: it takes an input list in almost any shape - a NumPy array, a list of integers, a list of floats, even a nested structure - and hands you back a clean Python list of floats. It exists because ComfyUI is loose with list types and strict nodes are not.

If you've done any animation work in ComfyUI you've hit the wall: one node hands you a numpy array, another hands you a list of ints, and a third refuses to accept either because it wants floats. That mismatch is exactly where workflows die with confusing errors. This node is the adapter that makes them agree. It also catches a subtler problem - a schedule of integers (say from a range node) will behave differently downstream than a schedule of floats, because a float list can carry fractional easing values while an int list can't.

How it works

The implementation is deliberately simple: it coerces the input through NumPy (converting dtype to float64) and returns a list. If the input is already a float list, it passes through untouched. There's no scaling, no rounding, no magic - which is the right behavior for a conversion node. One thing to note: it does not flatten nested lists or handle arrays of arrays the way you might hope; it's built for flat per-frame value lists, which is what the rest of this pack produces and consumes.

Inputs and output

• input_list - a wildcard input (*), so any list-like thing plugs in: FLOAT lists, INT lists, NumPy arrays, etc.
• Output is a FLOAT list.

That's the entire API. Wire whatever you have in, get floats out.

Installing it

Part of akatz-ai/ComfyUI-AKatz-Nodes. Install via ComfyUI Manager (search "AKatz") or:

cd ComfyUI/custom_nodes
git clone https://github.com/akatz-ai/ComfyUI-AKatz-Nodes
cd ComfyUI-AKatz-Nodes
pip install -r requirements.txt

Restart ComfyUI. No model downloads; requirements are numpy, torch, opencv-python and pydub.

Common issues

The realistic failure is feeding it a multi-dimensional array and expecting a flat list - it converts what it's given, so check what you're feeding if the result looks wrong. Otherwise this node has almost no failure modes, which is exactly why it's worth keeping on hand. Pair it with Adjust List Size when the type and length both need fixing in one shot.